HowTo: Get credentials from a cloud provider for GeneXus Cognitive API

This article explains how to get the configuration credentials needed to use the GeneXusAI Module.

Step 1 - Access to provider console

Login on your AI provider with your credentials (with previous registration).

• Alibaba Aliyun console
• Amazon Web Services console
• Baidu Cloud Environment console
• Google Cloud AI console
• IBM Bluemix console
• Microsoft Azure portal
• SAP Leonardo functional services (sandbox) or SAP Cockpit console (trial/production)
• Tencent AI console
• Google Cloud AI console (enable ML Kit API by Firebase)

Step 2 - Select the AI service

After login, you will be redirected to the developer console. Then, from the dashboard, create/enable a new resource for your desired AI task (you can filter services by name). This action usually asks you for a resource name, a region, and other general settings for the service you are creating (such information depends on the provider you use).

Regions supported by provider

If you do not set the region properly a GXAI5000 error will be raised with an 'Authentication Failed' message

Credentials by task

The following table summarizes which service credentials you need (by name) for each task in GeneXusAI. For detailed information, refer to the procedure description page.

Step 3 - Generate and set the credentials

Once your service has been created, the provider shows you the credentials you need. Also, these credentials will be located in a "Resource Management" section (or similar) in the provider's webpage in case you want to retrieve them again or regenerate them when it is needed. Once you get the credentials values, you simply set it properly on the corresponding Provider.Property field.

For example, for an API-key scenario, you can write a code like this:

&provider = new()

&provider.Name = !"{your_environment_name}"
&provider.Type = ProviderType.[Amazon|Baidu|Google|IBM|Microsoft|MLKit|SAP|Tencent]

&providerProperty= new()
&providerProperty.key = GeneXusAI.Configuration.PropertyKey.Key
&providerProperty.Value = !"{your_api_key_value}" // LINE (*)
&provider.Properties.Add(&providerProperty)

Please note that "{your_api_key_value}" value should not be pasted into your production code. As a best practice, you should read it from a configuration file or environment variable. For example, you can create a JSON file called 'provider.settings.json' with the following structure:

{
    "name":"{your_provider_name}",
    "key":"{your_api_key_value}"
}

Then, you can attach this file to your Knowledge Base as a File object, set the extraction properties (for copying it to the appropriate location) and read the credentials as follows:

&settingFile.Source = !"path/to/provider.settings.json"
&jsonString = &file.ReadAllText()       // read configuration file
&jsonProperties.FromJson(&jsonString)   // parse json structure
&myApiKey = &jsonProperties.Get(!"key") // get "{your_api_key_value}" value

Finally, you can replace LINE (*) with the following sentence:

&providerProperty.Value = &myApiKey

Step 4 - Execute your AI task

You will be able to execute your AI task!

Notes:

• For SAP Leonardo, Sandbox key can be retrieved on each API service by the "Show API Key" option and will be the same no matter what service you choose. For such reason, the documentation does not specify which service you should use (e.g. Inference Service for Optical Character Recognition (OCR) if you use GeneXusAI.Image.OCR task) due to the API key will be the same for every service.
• For IBM Watson, when you create a new resource, ensure to choose 'US South' region. Those services that allow authentication by username/password are maintained for backward compatibility (new instances will give you a key, not a username/password).
• For Google Cloud AI, you must create a single project and enable each API you desire to use. Then, on the "Credentials" option, you will able to create an API key for all those APIs you enabled.

Availability

• This document applies since GeneXus 16.
• Google Cloud AI provider is available since GeneXus 16 upgrade 1.